DOCUMENT RESUME ED 334 560 AUTHOR Nagy, William · PDF fileDOCUMENT RESUME ED 334 560 CS 010...

DOCUMENT RESUME

ED 334 560 CS 010 652

AUTHOR Nagy, William E.; And OthersTITLE The Development of Knowledge of Derivational

Suffixes. Technical Report No. 536.INSTITUTION Bolt, Beranek and Newman, Inc., Cambridge, Mass.;

Illinois Univ., Urbana. Center for the Study ofReading.

SPONS AGENCY Office of Educational Research and Improvement (ED),Washington, DC.

PUB DATE Aug 91CONTRACT 00087-C1001-90NOTE 32p.PUB TYPE Reports - Research/Technical (143)

EDRS PRICE MF01/PCO2 Plus Postage.DESCRIPTORS Elementary School Students; Intermediate Grades;

*Morphology (Languages); Reading Research; *ReadingSkills; Secondary Education; Secondary SchoolStudents; Skill Development; *Suffixes; Word StudySkills

IDENTIFIERS Word Structure

ABSTRACT

Knowledge of morphology--the ability to gaininformation about the meaning, pronunciation, and part of speech ofnew words from their prefixes, roots, and suffixes--is an importantcomponent of skilled reading. As part of a larger program of researchon morphology, a study explored the development of students,knowledge of common English suffixes. Two aspects of knowledge aboutsuffixes were investigated: (1) the ability to use the suffix on anew word to determine that word's part of speech and its appropriateuse in a sentence; and (2) the ability to distinguish between truesuffixed words (e.g., "swimmer") and nonsuffixed words that containthe orthographic form (but not the meaning) of a root word and suffix(e.g., "mother"). A battery of tasks measuring knowledge ofmorphology was administered to 720 fourth-grade, seventh-grade, andhigh school students. Knowledge of common English suffixes was foundto continue to develop after fourth grade; even in high school, somestudents had serious problems with these suffixes. Knowledge ofmorphology appears to be a distinct component of verbal ability,although it is significantly related to standardized measures ofreading ability in seventh grade and high school. The instrumentsdeveloped in the study identified students who have particulardifficulties with English suffixes, and thus have potential asdiagnostic tools. (Author)

**************************************g********************************Reproductions supplied by EDRS are the best that can be made

from the original document.***********************************************************************

CENTER FOR THE STUDY OF READING

/2126Technical Report No. 536tkirs

THE DEVELOPMENT OF KNOWLEDGEOF DERIVATIONAL SUFFIXES

William E. Nag, Irene-Anna N. Diakidoy,and

Richard C. AddemonUniversity of Illinois at Urbana-Champaign

August 1991

University of Illinois at Urbana-Champaignil Gerty Drive

Champaign, Illinois 61820

BEST COPY AVAILABLE

U.11. DEPARTMENT OF EDUCATIONOffice Or Educational Rasura and improvementEDUCATIONAL RESOURCES INFORMATION

CENTER (ERIC)*This doCument has Peen rePrOduCdreceived from the pinion or orgenitationoriginating itr Minor Changes have peen made to ImproversproduCtOn Quality

Points of view or opinions slated in this docu-ment do not neCeMarity represent &twistOERI position or polity

The work upon which this publication was based was supported in part by theOffice of Educational Research and Improvement under Cooperative AgreementNo. G0087-C1001-90 with the Reading Research and Education Center. Thepublication does not necessarily reflect the views of the agency supporting theresearch.

James Armstrong

Gerald Arnold

Diana Beck

Yahaya Bello

Diane Bottom ley

Clark A. Chinn

Candace Clark

John Consalvi

Iren e-Anna N. Diakidoy

Colleen P. Gilrane

Barbara J. Hancin

Richard Benne

Michael J. Jacobson

EDITORIAL ADVISORY BOARD1990-91

MANAGING EDITORFran Lehr

MANUSCRIPT PRODUCMON ASSISTANTSDelores Plowman

Debra Gough

3

Carole Janisch

Bonnie M. Kerr

Paul W. Kerr

Daniel Matthews

Kathy Meyer Reimer

Montserrat Mir

Jane Montes

Juan Moran

Anne Stallman

Bryan Thalhammer

Marty Waggoner

Ian Wilkinson

Hwajin Yi

4,

4

Nagy, Diakidoy, & Anderson Derivational Suffixes - 1

Abstract

Knowledge of morphology--the ability to gain information about the meaning, pronunciation, and partof speech of new words from their prefixes, roots, and suffixesis an important component of skilledreading. As part of a larger program of research on morphology, a study explored the developwent ofstudents' knowledge of common English suffixes. Two aspects of knowledge about suffutes wereinvestigated: The ability to use the suffix on a new word to determine that word's part of speech andits appropriate use in a sentence, and the ability to distinguish between true suffixed words (e.g.,swimmer) and nonsuffixed words that contain the orthographic form (but not the meaning) of a rootword and suffix (e.g., mother). A battery of tasks measuring knowledge of morphology was administeredto 720 fourth-grade, seventh-grade, and high school students. Knowledge me. common English suffixeswas found to continue to develop after fourth grade; even in high school, some students had seriousproblems with these suffixes. Knowledge of morphology appears to be a distinct component of verbalability, although it is significantly related to standardized measures of reading ability in seventh gradeand high school. The instruments developed in the study identified students who have particulardifficulties with English suffuces, and thus have potential as diagnostic tools.


THE DEVELOPMENT OF ICNOWLEDGE OF DERIVATIONAL SUFFIXES

Skilled reading depends not just on knowing many words but on being able to deal effectively with newones. The nature of our language is such that one cannot expect to have prior knowledge of all thewords one will encounter in a text. Nagy and Anderson (1984) estimated that the average fifth grader,in the course of a year's reading, may encounter more than 10,000 new words, that is, words the studenthas not previously encountered in print. It is clear that no program of vocabulary instruction, howeverambitious, could attempt to cover all the new words a student will read. What students need, then, isnot just knowledge of many words (although this is certainly helpful), but strategies for handling newwords in text (Mason, Herman, & Au, 1990).

In the initial stages of learning to read, most of the new words students encounter in print are alreadyin their oral vocabularies. For this type of new word, the student can rely on knowledge of spelling-sound correspondences. After third grade, on the other hand, students encounter increasing numbersof words in print that are not in their oral vocabularies; decoding strategies alone will not provideinformation about their meaning. However, tbe majority of new words in text are related to morefamiliar ones through grefixation, suffocation, or compounding (Nagy & Anderson, 1984; White, Power,& White, 1989). In the course of a year's reading, a reader will encounter numerous words such asuncourteous6,, queenlike, or playfellow. These are long, rarely occurring words; yet they are notnecessarily difficult words, if the reader can recognize the familiar parts and understand how these partsfit together. Hence, students' knowledge of morphologythat is, the ability to gain information aboutthe meaning, pronunciation, and part oc speech of new words from their prefixes, roots, and suffixes--determines in part how effectively they will be able to deal with new, long words.

It is well icnown that word length is associated with text difficulty (Davison & Green, 1988; Klare, 1984).Howevei, it has been argued that word length per se is not a cause of difficulty for all readers(Anderson & Davison, 1988). The vast majority of long words have some discernable morphologicalstructure; it is'difficult to find a word of mon, than eight letters that does not contain some recognizableprefix, stem, or suffix that gives at least partial information about its maning and pronunciation.Anderson and Davison hypothesize that word length is a source of diffiaty for those younger and lessable readers who cannot analyze a long word such as sleeplessness into its component parts. Thishypothesis is consistent with Baker's (1989) fmdings that third graders were more likely to identifylonger rather than shorter nonwords as problematic, whcreas fifth graders were not influenced by wordlength. Word length influences third graders' judgments of the difficulty of reading the words inisolation, and also of the difficulty of understanding sentences containing he words. The particulaidifficulty that long words pose for younger and less able readers was also confirmed by a reanalysis ofdata reported in Nagy, Anderson, and Herman (1987), which showed that word length had anindepeLdent negative effect on learning Lorn text only for students in the bottom third of a sample ofmiddle-grade readers.

Ju.. and Carpenter (1987) argued that if the burden of decoding a long word is too great, readers willnot have sufficient capacity in working memory to store and process the entire sentence. AlthoughBaker (1989) did not examine the possible role of morphological knowledge in the difference betweenthird and fifth graders, it is reasonable to assume that the ability to chunk the letters in a long word intomeaningful morphemes would facilitate the processing of new words, and that this ability increasesbetween third and fifth grades. Hence, differential knowledge of morphology may account, at least inpart, for the relationship between word length and text difficulty, and for the problems that long wordspose for younger and less able readen.

Considerations such as the above have served to highlight morphological knowledge as an importantcomponent of vocabulary growth and skilled reading. In addition, there has been shown to be arelationship between general verbal ability and the use of morphology in learning new words (Freyd &Baron, 1982) and in the comprehension of sentences containing suffixed words (Tyler & Nagy, 1990).

5


The potential benefits of instruction in morphology are considerable (White, Power, & White, 1989).The need for instruction is also clear; upper-elementary and middle-school students ...Io not appear tomake anything close to full use of the help that morphological relatedness among words should providein word learning (Freyd & Baron, 1982; Wysocki, & Jenkins, 1987). Morphology does receive a fairamount of attention in most reading programs, generally under the label "structural analysis" (Winsor,Nagy, Osborn, & O'Flahavan, in preparation). What is not clear is the quality or effectiveness on suchinstruction.

Current instructional practice operates largely in s theoretical vacuum. There has been little researchdone to test whether instruction on word parts has any effect (see Graves, 1986), and none that we knowof on the relative effectiveness of different instructional approaches. One reason for the paucity ofinstructional research on morphology is the lack of a foundation in basic research. Any attempt toformulate an effective method of irstruction must be based on accurate information about what studentsalready know about morphology, and how they use this knowledge in reading and word learning.

Aspects of Morphological Knowledge

Since Berko's (1958) pioneering study, a number of researchers have examined children's acquisitionof knowledge about morphology (e.g., Condry, 1979; Denying & Baker, 1979; P:eyd & Baron, 1982;Selby, 1972; Shepherd, 1973; Sternberg & Powell, 193; Tyler & Nagy, 1989; 1990; Wysocs.:. & Jenkins,1987). For there to be a solid research basis for making decisions about instructional practice, however,progreu is nece,sary on two fronts. First, there is still need for more detailed information about whatchildren knqw about morphology, when they learn it, and how they use it. Second, a framework mustbe developed that allows the diverse and sometimes seemingly inconsistent information about children'sacquisition of morphology to be put together into a coherent picture. Such a framework mustdistinguish between, and explicate the relationships among, different aspects of morphologicalknowledge.

Different Types of Word Parts

An es ,ential beginning step in gaining an understanding of childrni's acquisition of morphologicalknowledge is recognition that s"..cb knowledge is not monolithic. We will, therefore, lay out here someof the distinctions that must be recogniml in gaining a comprehensive picture of this topic.

First, one must distinguish between different kinds of morphological structures--between prefixation,suffixation, and compounding, and between inflectional and derivational suffixes. Inflectional suffixesare those that create different forms of the same word, ,or example, the endings for tense (walk, walked,walking) or number (frog frogs). Derivational suffixeF create different woras, typically with differentparts of speech (refine, reflnemern). Children pretty well master inflections in oral language hi theirpreschool years (Berko, 1958). They largely learn other types of word formation processes later(Derwing & Baker, 1979).

Although it is difficult t3 rank compounding, prefixatiork and suffixation in any absolute scale ofdifficulty or age of acquis ition, it is clear that these are qualitadvely different processes. One dimensionalong which they differ is the meaningfulness of their parts. Compounds typically consist of elementsthat are words in their own right. Prefixes and suffixes are bound morphemes; they are not words inthemselves, and their meanings or grammatical functions are often difficult to articulate (it is not easy,for example, to express in simple terms what the a in asleep or the ive in progressive means). The simplefact that prefixes come at tlac beginning of words and suffixes at the end may also have flipUcations forhow readers process these word parts.

Another important distinction that must be made is between bound and free stems. Free stems, suchas the teach in teacher, are words in themselves Bound stems are stems such as the proach in reproach


and approach, the ceive in receive, deceive, and perceive, or the petro in petrochemical, which do notappear in isolation. Shepherd (1973) found that unlike free stems, bound stems appeared to play littlerole in the vocabulary knowledge of college students. Presumably they would play still leu of a role foryounger students. Some studies, however, have shown a measurable increase in knowledge and use ofbound stems between high school and college (Nagy & Scott, 1990; Sternberg & Powell, 1983).

Among bound stems, one might want to make a further distinction between Latin roots such as ceive

or fer that are almost entirely void of meaning in current English, and Greek roots such as geo, thenno,

or graph, which often make a discernable contribution to the meaning of the words in which they occur.

Different Types of Knowledge about Word Structure

Thus, there are several different types of word structure, which may function somewhat differently inthe language. One cannot assume that knowledge of one type would automatically transfer to the other.One must also distinguish between different types of knowledge about word structure. Four differenttypes of knowledge about word structure can be identified:

1. Recognition of morphological relatedness. Perhaps the most basic level of knowledge aboutmorphology is the recognition that two words are reldiedfor example, seeing the thanks inThanksgiving. This basic insight presumably develops relatively early. Tyler and Nagy (1989) found nogain in the ability to recognize that a novel word contains a known stem after fourth grade.

2. The ability to distinguish stems from pseudo-stems. The child who learns to see the agree inagreement might also find the sea in season. The ability to distinguish between these two kinds of casesis a second aspect of morphological knowledge, one that presumably grows as thc child begins toappreciate the systematic nature of morphological relationships among words. Derwing and Baker(1979) found that young children were more likely to rely on tb3 spelling of a word, relative to itsmeaning, than adults were, in making judgments about relationships among words. Instructionencouraging students to look for little words in big words" may encourage this kind of confusion. Anyinstructional approach to morphology must make some provision for dealing with this possiblemisconception.

3. Knowledge of individual word parts. A child may wei! learn to recognize -ness as an English suffixbefore -ship. One type of growth in knowledge of morphology is therefore the learning of particularprefixes, roots, and suffixes. Much of the growth in morphological knowledge between high schoel andcollege found by Nagy and Scott (1990) and Sternberg and Powell (1983) might be attributed to growthin knowledge of particular stems and affixes.

4. Knowledge of structural meaning. Another aspect of what students know about morphology isstructural knowledge, that is, knowledge not just of ihe meanings of individual word parts, but how themeanings .1 these parts fit together to make up the meaning of the whole word. There is a variety ofevidenc,k to suggest that this type of knowledge is relatively late to develop, and strongly related to verbalabir ty.

Gleitman and Gleitman's (1970) research on the ability of adults to interpret complex (i.e., three-part)compounds illustrates this point. Compounding appears to be one of the first word formation processesto be acquired; children are certainly able to understand, and to generate, novel compounds before theyenter school (Clark, Gehnan, & Law. 1985; Derwing & Baker, 1979). However, Gleitman andGleitmaa found substantial differences associated with verbal ability and educational background inadults' ability to interpret the meaning of novel complex compounds. In their experiment, the correctinterpretation of the compounds depended not on recognizing the component parts, but on determiningtheit relationship. For example, the subjects would have to decide wh3ther abird house foot was morellkely to mean the foot belonging to a house for birds, or the foot of a bird who lived in a house. What

7


is striking about these results is the existence of ability related differences, among adults, in theknowledge of structural meaning for a linguistic structure that must have been acquired much earlier.(To some extent, the interpretation of complex compounds depends on knowledge about the way thingsare in the world (Kuhara-Kojma & Hatano, 1991]. However, the differences in ability to interpretcompounds that Gleitman and Gleitman (1970] found could not be attributed to any large extent todifference in background knowledge; rather, they seem to reflect differential ability to utilize cuesprovided by linguistic structure.)

Likewise, there is evidence that knowledge of the structural meaning of derivational suffbces is not welldeveloped among school-age children. Investigating students' ability to define simple and suffixed words,Freyd and Baron (1982) found that all the students they tested, including the high-ability group, tendedto ignore or misinterpret the suffix when defining a word; that is, if asked to provide a defmition for theword expansive, they would generally provide a definition that would be appropriate for expand.Wysocki and Jenkins (1987) obtained similar results; when fourth-, sigh-, and eighth-grade studentswereasked to provide a definition of a derivative, they most frequently responded with the definition for itsstem.

Tyler and Nagy (1989) investigated the relative order of acquisition of different aspects of knowledgeabout derivational suffixes. As mentioned above, they found no increase in students' basic recognitionof morphological relatednessthe ability to see the familiar stem butter in the novel derivative butteriess--after fourth grade. They also attempted to measure students' knowledge of the syntactic function ofsuffixes, that is, what the suffix contributes to the meaning of the derivative as a whole, and foundsubstantial growth in this knowledge after fourth grade.

Tyler and Nagy (1990) investigated the relative contribution of these two aspects of morphologicalknowledge to high school students' comprehension of complex sentences containing suffixed words.They found that high school students did not differ substantially in their recognition of morphologicalrelatedness, but did find a significant relationship between general verbal ability and the students' useof the syntactic informatinn provided by the sulfa.

Effects of' Task and Stem Familiarity

The extent to which students demonstrate knowleibe of morphology is likely to depend on the natureof the task they are asked to perform. One distinction that must obviously be made is that betweenrecognitim. and production tasks. Another task dimension that may influence students' ability todemonstrate knowledge of morphology is the familiarity of the stems involved. When the task involvedseeing the relationship between a derivative and a familiar stem (e.g., seeing the agree in agreement orthe butter in butteriess), secoad graders (Condry, 1979) and fourth graders (Tyler & Nagy, 1989) werefowid to perform quite adequately. On the other hand, in the task posed by Freyd and Baron, eighthgraders failed to perceive or utilize the relationship between novel stems and derivatives (e.g., skaf*steal," skew "thief').

Tyler and Nagy (1989) directly compared students' ability to utilize the syntactic information in suffixeswith two types of words: possibly familiar words with known stems, and suffixed words with nonsensestems. 'PI; task, illustrated by the item in Table 1, was to choose which suffixed word best fit in asentence context. Strictly speaking, this task required no knowledge of the meaning of the stem; thecorrect answer depends entirely on the suffix. However, students consistently performed better on thereal-word version than on the nonsense-Ltem version.

[Insert Table 1 about here.]

There are two possible reasons for this difference in performance. One is that some students did not,in fact, know the syntactic value of the suffixes. They were able to answer correctly on the real-word


items because they knew the particular words involved, as unanalyzed wholes. That is, they knew(implicitly at least) that intensify is a verb (and hence fits the sentence context) and that intensiry is a

noun (and hence does not) in the same way that they know rejoice to be a verb, and Joy to be a noun,

without any help from a suffix.

The other is that students may have had some knowledge of suffixes, but were less able to apply thisknowledge in the case of the nonsense-stem items, perhaps because of the additional demands of tryingto choose among nonsense words. These two accounts are, of course, not mutually exclusive.

Study Goals

The research reported here investigated students' acquisition of familiar English derivational suffixesbetween fourth gade and high school. In particular, it focused on the syntactic function of thesesuffixeshow the suffix affect.s the meaning, and more specifically, the part of speech, of the suffixed

word.

There are several reasons why we chose this particular topic. Most important, this appears to be a late-developing aspect of knowledge about suffixes. Both Freyd and Baron (1982) and Wysocki and Jenkins(1987) found that students as old as eighth graders showed no apparent knowledge of what a suffixcontributed to the suffixed word. Tyler and Nagy (1989) did not find consistent evidence that fourth-grade students possessed this type of knowledge, but did find it in sixth graders and older students.Even in high school, Tyler and Nagy (1990) found significant ability-related differences in students'tendency to utilize the syntae.ic information provided by suffixes.

Another reason for looking at students' knowledge of the syntactic function of suffixes is that thisknowledge is important for writing. Suffixed words are more common in written than in oral language(see Chafe & Danielewicz, 1987), and the writer must have a gasp of how the suffixes work in orderto use these words correctly.

It is likely that the studies by Freyd and Baron (1982) and Wysocki and Jenkins (19871 underestimatedstudents' knowledge of the syntactic function of suffixes (as Wysocld and Jenkins lloted), becausestudents' knowledge was measured in terms of their ability to provide definitions for derived words.Providing an adequate definition for a derived word, say an abstract noun like cancellation, or anadjective like responsive, is difficult enough for lexicographers, let alone for students. Abstract noundefinitions often resort to phrases such as "the act or process of," and derived adjectives are sometimesgiven definitions starting with "of or pertaining to." It is not surprising then, that elementary and middleschool ctudents were not able to articulate dermitions of derived words that expressed their part ofspeech. The type of items used by Tyler and Nagy (1989) to measure students' knowledge of thesyntactic function of suffixes avoided this problem, but, as we will discuss later, such items have otherweaknesses. One of these was a confounding of knowledge of suffixes with other aspects of readingability.

Our primary goal in the present study, therefore, was to develop a paper-and-pencil task that wouldenable us to accurately measure students' knowledge of the syntactic function of suffixes, in a way thatwould allow us to assess the relationship between this knowledge and measures of general verbal ability

in a meaningful way.

The suffixes we investigated were chosen from among the most frequently occurring suffixes in English(using data from Harwood and Wright, 1956). We also limited our choice of suffixes to neutral suffixes

(see Tyler & Nagy, 1989), that is, suffixes that are added only to free-standing words, and that make onlyminimal changes in the spelling and meaning of the words to which they are added. Thesesuffixes werechosen so that we could be relatively sure that errors by students reflected problems with using thesyntactic information providtd by the suffix, rather than lack of recognition of the suffix.

9


We were also interested in finding out whether students' knowledge of the syntactic function of suffixesa related to other aspects of morphological knowledge. We therefore included in this study aquestionnaire to assess students' self-reported reliance on morphology (and other word-level strategies)when they encountered new words and a test designed to measure students' ability to distinguishbetween true roots (e.g., the swim in swinuner) and pseudoroots (e.g., the moth in mother).

Method

Subjects

Subjects were 700 students from schools in middle-sized midwestern towns--236 fourth graders from fourelementary schooLs, 125 seventh graders from two junior high schools, and 339 students from two highschools. There were 120 tenth-grade students from one high school (High School A). Students fromanother high school (High School B) were drawn from 12 reading and literature classes selected toprovide a variety of ability levels; each class contained students from different grade levels. Altogether,among the 219 subjects from High School B, there were 70 students in ninth grade, 81 in tenth, 39 ineleventh, and 29 in twelfth.

Different standardized test scores were available for different grade levels, and for the two high schools.Thus, it is not possible to determine fully how much of the differences between grade levels might reflecttruly developmental effects and how much they might be due to differences in the make-up of thecla,ses tbat were tested.

Suffix Function Test

The focus of this study was students' knowledge of the syntactic function of derivational suffuses. Sincederivational suffixes serve primarily to mark changes in part of speech (e.g., happylhappiness), weconstructed a test to measure students' knowledge of how these suffuses affected use of the suffused wordin sentences.

We made several improvements on the items Tyler and Nagy (1989) used to measure students'knowledge of the syntactic function of suffuses. The intent was to provide an item that could beanswered correctly only if one recognized the syntactic function of the suffixes. The first item in Table1 fulfills this requirement, with one qualification: A person might know the part of speech of each ofthe choices (intensify, intensification, intensity, intensive), but only know these words as unanalyzableunits. That is, the person might know these words without any knowledge of what the suffix contributesto the part of speech, in the same way a person knows the part of speech of the words joy,friend, sick,and think.

Tyler and Nagy employed a second type of item to circumvent this problem. This type of item usedsuffixed words with nonsense stems. Therefore, it was only by knowing the part of speech conveyed bythe suffix that a person could get the item right. However, a person might fail to get such an item rightfor other reasons, in particular, because the presence of a nonsense stem might be confusing. Hence,A person might get an item of the first type righ without knowing anything about the part of speechconveyed by a suffix, and fail to get an item of the second type right even if he or she had someknowledge of the function of the suffix.

Another problem with these items is that the persoies ability to answer the item reflects not only aknowledge of suffuses, but an ability to read and interpret the seLtence context.

For the present study, we co....,tructed a new type of measure to assess students' knowledge of thesyntactic properties of suffixes. We used novel (or very low-frequency) but well-formed suffused wordswith high-frequency stems. For example, the word butteriess is extremely rare, so students would be


unlikely to have encountered it before; if they recognize it, it is because they analyze it into a familiarstem plus a suffix. On the other hand, the stem itself is familiar.

For each suffixed word a multiple-choice item was created to assess the students' knowledge of how thesuffix governed the use of that word in the sentence. We will refer to these as "derivative items." Table2a gives a sample derivative item. For each such item, a corresponding "stem item" was constructedusing a frequent, non-suffixed word of the same part of speech that made sense in the same sentencecontexts. Table 2b gives a sample stem item.

[Insert Table 2a and Table 2b about hem.]

The Derivative version of the item measures the students' ability to determine the syntactic function ofa novel suffixed word. The Stem version of the item measures everything else that contributes to astudents' performance on such an itemability to understand the context sentence, test taking skills,possible effects of the order of choices, and so on. Thus, the Stem version of the item provides us witha baseline against which to evaluate students' performance on the Derivative version. The onlydifference between the two types of items is that the Derivative items require specifically morphologicalknowledge. When Stem-item performance is controlled for statistically, students' performance onDerivative items gives us a rather pure measure of their ability to use their knowledge about suffixesto figure out the part of speech of a novel suffixed word.

If a student were able to make full use of the information provided by suffixes, the score on Derivativeitems itould be as high as that on Stem items. In fact, one might even expect the score for Derivativeitems to be higher; Although powderize is a novel word, it is explicitly marked by the suffix for part ofspeech--it is ibviously a verb. On the other hand, most non-suffixed English words are at leastpotentially ambiguous as to their part of speech. Smash, for example, is typically a verb, but it issometimes used as an adjective (a smash hit), and could conceivably be used as a noun (it was a realsmash/the smash of the pumpkin on the sidewalk). To the extent that students score lower onDerivative items than Stem items, then, this can be taken as an indication that they are failing to usethe syntactic cues provided by the suffixes.

Twenty rare sufruced words were selected; rescuable, powderize, marer, echoful, activeness, weaponist,parentism, butteriess, cheesish, cutely, mirrorize, baricless, orangeness, frcggish, snal.lable, evilly, greasefu4profttism, herdist, and repairer. The stems of these words range in frequency from 8.8 to 37.2occurrences per million words of text, and can be assumed to be familiar to most students. The suffixeswere also chosen to represent common English suffixes. However, the particular stem +suffixcombinatioas are very rare: Only two of the suffixed words themselves occurred in Carroll, Davies, &Richman (1971), evilly and roarer. These two words are found very infrequently, occurring less thanonce in 10 million words of text. It was assumed that most students would not previously have seen anyof the suffixed words as units, and would therefore have to analyze them into stem and suffix tounderstand them.

Two versions of a multiple-choice test were constructed, with each version containing 10 derivative and10 stem items in a random order. Any given subject saw only one versiun of an item.

Stem Discrimination Test

Another aspect of knowledge of suffixes is reflected in students' ability to discriminate between stemsand pseudostems--that is, to recognize that there is not a moth in mother, at least not in the same waythat there is a bath in bather. It is likely that at some point in their acquisition of concepts like rootword, suffir, and prefix, some students develop misconceptions, and do not distinguish true rootwordsfrom pseudostems.

1 1


A test was constructed to measure students' ability to discriminate stems from pseudostems. Studentswere given a brief explanation of the concept "root word," emphasizing the distinction between rootwords and simply finding little words in big words (such as the fai in father). They were then instructedte circle all the root words they could find and to leave words without roots uncircled.

The test consisted of 64 words, printed in two columns on a page, double spaced and with spacesbetween the letters to facilitate circling. Half of the words contained genuine root words, and half didnot. Of the 32 words without true root words, 12 had no apparent roots at all (common, simple, guess)and 20 had pseudoroots (needle, season, corner). (We had originally thought that some of the wordsin this second group belonged to the first group, but students surprised us by circling the had in shade,the me in meal, and the kit in kitchen, which we had not originally seen). Of the 32 words that didcontain genuine root words, 17 were common suffixed words (thinly, selfish, wonderful), and 15 werenovel suffixed words with familiar stems (jumpsome, bounceful, clownish).

Three versions of this test were constructed, differing only in the order of the items.

New Word Strategy Questionnaire

A questionnaire was constructed to obtain self-report measures about what students do when theyencounter new words. The questions included are listed in Table 3.

[Insert Table 3 abo.ot ;?ere.]

For the first question, the possible choices were (a) about one in every sentence, (b) a few in everyparagraph, (c) two or three on a page, (d) one or two a day, and (e) I know almost every word I read.For the remaining questions, the choices were (a) always, (b) most of the time, (c) about half the time,(d) once in a while, (e) almost never, and (f) never.

As self-report data, the answers to questions such as these would not necessarily be a very accuratereflection of what students do, or how often they do it, when they encounter unfamiliar words. On theother hand, such a questionnaire does provide some information about what students think they do, andabout the relative reliance they think they place on different strategies.

Standardized Measures of Reading Ability

For most subjects, standardized test scores for reading comprehension and vocabulary were obtainedfrom school files.

For some fourth graders, we were able to obtain Reading Total scores on the SRA Level 3S, Form 0,expressed in terms of national percentiles. These tests had been administered the preceding year.Other fourth graders were in schools that would not release information on students' test scores.

For seventh graders, we obtained Reading Comprehension scores on the Stanford 7 Plus test, againexpressed in terms of national percentiles.

For High School A, we obtained scores from the Stanford 7 Plus test, Reading Comprehension,expreued in terms of national percentiles. For High School B, we were able to obtain students scoresfrom the Stan!' f d Diagnostic Reading Test (Blue level). The scores we used were RladingComprehension (total), Vocabulary, Word Parts, and Structural Analysis. These were raw scores.

The Reading Comprehension (total) score was the sum of scores from two subtests that measured literaland inferential comprehension. The Word Parts subtest required students to choose the best synonym

1 2


for an underlined word part. For example, in the word caulk the student might have to choosebetween (a) throw (b) burn (c) hinder (d) cover.

The Structural Analysis subtest items consisted of four groups of one or more letters, three of whichcould be reordered and assembled to form a word. However, for some items, no word could be formed.The student was to answer by choosing the one part that did not fit, or else to respond "no word." Theword parts were not morphemes, that is, not meaningful parts of words. For example, an item mightlook like this:

a) skb) bac) add) ete) no word

Results and Discussion

Sutra Function Test

Means for students' performance on the Suffix Function test are given in Table 4. As ca.n be seen fromthe table, subjects did better on the Stem version of items than on the Derivative version, meaning thatthey were not utilizing all the information provided by the derivational suffixes. The gap between theStem and Derivative versions is greater in fourth grade than in the higher grades, suggesting that fourthgraders have acquired somewhat less knowledge of suffixes than the older students. On the other hand,at all grade levels, subjects performed significantly above chance on the Derivative version, showing thatthey do have substantial knowledge of the syntactic properties of common suffixes.


Another way of representing the results from the Suffix Function test is in terms of a scauerplotcomparing students' scores on the Derivative items with their scores on the Stem items. Such ascatterplot is given in Figure 1. The Y-axis represents students' proportion correct for Derivative items,corrected for guessing. (Scores less than zero have been set equal to zero.) The X-axis representsstudents' proportion correct on Stem items, also corrected for guessing. The scatterplot in Figure 1includes data from all students at all grade levels. Preliminary scatterplots done separately for fourthgrade, seventh grade, and high school all showed the same pattern.

[Insert Figure 1 about here.]

It is clear from Figure 1 that performance on Stem and Derivative items is correlated, r .67,p < .001.(Correlations for individual grade leveLs are about the same--for fourth grade, r .59, for seventh grade,.59, and for high school, .70.) However, the figure tells us more than just the fact that these twovariables are correlated. What is clear, beyond the correlation, is that there are very few students abovethe diagonal, but quite a few below it.

The lack of points above the diagonal is not surprising; to be above the diagonal, a student would haveto have done better on the Derivative items than on the Stem items. Since the Derivative items measureknowledge of the syntactic properties of suffixes plus all the knowledge and skills required for doing theStein items, this is expected.

The relatively large number of points below the diagonal shows that ability to answer the Stem itemsmay be a necessay condition for answering the Derivative items, but it is not a sufficient condition.There are many students who demonstrate a high level of facility with the Stem items, but who are not

1 3

Nag, Diakidoy, & Anderson Derivational Suffixes - 11

able to demonstrate any knowledge of suffixes. This indicates that some subjects have a specificproblem with suffixes.

Students' performance on the Derivative version of items on the Syntactic Function test serves as ameasure of their knowledge of the syntactic function of suffixes, when their performance on the Stemversion of items is used as a covariate. It turned out to be somewhat complicated to remove varianceassociated with performance on the stem items, because of curvilinearity of this measure, and becauseof slight differenres in difficulty of the stem items in the two versions of the test. Therefore, to bestmeasure subjects nowledge of the syntactic function of suffixes, we included in a regression equation(functioning as covariates) subjects' performance on the stem version of items, a dummy variablerepresenting test version, and the interaction of these last two variables. To deal with curvilinearity, wealso included the square of the subjects' stem version score.

This regression model was used to compute a residual difference score, representing in effect subjects'performr....ce on the Derivative items of the SyntacticFunction test, when their performance on the Stemitems has been wed as a covariate. We take this residual difference score, which we shall henceforthcall the Suffix Function score, to be the best measure of students' ability to we their knowledge aboutthe syntactic properties to predict the syntactic function of novel suffixed words.

It should be stressed here that by defining the Suffix Function score in this way, we have removed allvariance associated with test taking skills and with knowledge of the vocabulary and sentence structuresused in the context sentences of the Suffix Function test. This leaves us with a relatively pure measureof students' knowledge of the syntactic contribution of suffixes. It also means that correlations betweenthe Suffix Function score and standardized measures of reading ability represent a very conservativeestimate of the relationship between these coastructs.

For Table 4, Suffix Function scores were computed for all subjects using a single regression model, toallow for a comparison of students' knowledge of the syntactic function of suffixes across grade levels.One-way analysis of variance using the Suffix Function score as the dependent measure revealed thatthere was a significant difference among the three groups, F(2,629) 11.4, p < .001. Post hoc Tukeytests indicate that the only significant difference is between fourth grade and high school. For otheranalyses, Suffix Function scores were computed using the regression model for each group individually.

Stem Discrimination Test

Three variables were computed to represent subjects' performance on the Stem Discrimination test.

Proponion of true roots circled is the proportion of words with true roots for which the subject in factcircled the root. Proportion ofpseudoroots circled is the proportion of words with pseudoroots for whichthe subject circled the pseudoroot. (In the case of these first two variables, instances of other types oferroneous circling, for example, circling nonword subparts of words, were excluded from the calculation.)

These first two variables were found to be correlated, r a .22, p < .001. This suggests that the firstvariable reflects two componentsfirst, the ability to identify root words, and second, a tendency to circleword parts, whether true or pseudoroots. To help distinguish between these two factors, a thirdvariable, stem discrimination, was computed.

Stem discrimination is the proportion of circled word parts that were in fact true roots. (In calculatingthis proportion, other types of erroneous circling e.g., circling of nonword subparts of words, wereincluded in the denominator.) If a subject circled only true roots, this variable would have a value of1.0; if a subject circled true roots and pseudoroots equally often, it would have the value. .615 (sincethere were 20 possible pseudoroots to circle, and 32 true root words). If a subject were to circle only

Nagy, Diakidoy, & Anderson Derivatirmal SIffixes - 12

pseudoroots and no true roots the variable would have a value of zero. Means for these three variablesare given in Table 5.


These scores reveal that the majority of subjects were able to discriminate between true roots andpseudoroots. However, at all grade levels, there were some subjects who did not consistentlydiscriminate.

One-way analysis of variance showed a significant difference among the three groups of subjects forProportion of True Roots Circled, F(2,624) 32.6, p < .001, Proportion of Pseudoroots Circled,F(2,624) 4.9, p < .01, and Stem Discrimination, F(2,624) - 9.4, p < .0001. Post hoc Tukey testsrevealed that for Proportion of True Roots Circled, the mean for fourth grade was lower than the meansfor the other two groups. For Proportion of Pseudoroots Circled, seventh-grade students had a lowermean than the other two groups; for Stem discrimination, seventh-grade students had a higher meanthan the other two groups. The surprising svperiority of seventh graders over high school students mightbe due to overall differences in ability between these two groups. However, as will be discussed below,this effect probably reflects the conditions under which some of the high school students were testedrather than differences in the students' knowledge of morphology.

Knowledge of the Syntactic Function of Suffixes and Standardized Ability Measures

Table 6 gives the correlations for each group of subjects between standardized ability measures and thevariables from the Suffbc Function Test. Correlations were computed separately for High Schools A andB, because of differences in the standardized ability measures available for these two schools. Severalthings are apparent from this table. For one, it is clear that computing the Suffix Function Score usingthe Stem score as a covariate reduces, but does ne totally remove, the correlation between performanceon the Derivative items and standardized ability measures. For the seventh-grade subjects, and thosefrom High School B, there is a substantial correlation between the Suffix Function Score andstandardized ability measures. As noted above, because of the use of the Stem score as a covariate incomputing the Suffax Function Score, this correlation is a very conservative estimate of the relationshipbetween reading ability and students' knowledge of the syntactic function of suffixes.


For the students in High School B, we were able to obtain scores for several subtests of the StanfordDiagnostic Reading Test: The Reading Comprehension Total, Vocabulary, Word Parts, and StructuralAnalysis. As can be seen from Table 9, the Word Parts subtest was the one most strongly related tostudents' knowledge of the syntactic function of sufroms.

In subsidiary regression analyses, the effect of the Word Parts subtest was found to be significantwhether this variable was entered before, F 21.1,p < .001, or after, F 10.0,p < .01, students' rotalReading Comprehension score had been entered. The effect of the latter variable was also significant,F 10.6,p < .01, but only if it was entered before the Word Part subtest score. On the one hand, thisis Lot a surprising result: Both the de:ivative items in the current study and the Word Part subtest ofthe Stanford Diagnostic Reading Test purport to measure students' knowledge of word parts. On theother hand, there are several differences between the two tests. Of the 30 items on the Word Partsubtest, only one specifically involves a suffix. The nature of the tasks is also different; in the Word Partsubtest, students are asked to choose the answer closest in meaning to a particular underlined word part;there is no context sentence involved in the items, and part of speech plays no role. Our results show,however, that despite their differences, these two tasks appear to tap a similar ability to deal with wordparts, which cannot be reduced to general reading ability or vocabulary knowledge.

1 5

Nagy, Diakidoy, & Anderson Derivational Suffuces - 13

The Structural Analysis subtest, on the other hand, deals with students' ability to manipulate non-morphemic word parts. Hence it is not surprising that performance on this subtest is not as stronglyrelated to subjects' Suffix Function knowledge.

Stem Discrimination and Standardized Ability Measures

Table 7 gives the correlations between standardized ability measures and the three variables derivedfrom subjects' performance on the stem discrimination test.

[Insert Table 7 about here.)

As can be seen from the table, proportion of true roots circled is significantly related to ability only infourth grade. This suggests that for fourth graders, identifying the roots of derivatives is still developing;in later grades, subjects have reached a ceiling on this task. For older subjects, discriminating betweentrue roots and pseudoroots is still an issue, and hence highly related to ability measures.

The high correlation between Stem Discrimination and reading comprehension ability for seventhgraders is striking, especially considering that the seventh graders appear to be close to a ceiling on thestem discrimination task.

Summary of Correlations Among Measures

Table 8 gives a summary of correlations among the different measures explored in this study.


Students' Suffix Function scores (the residual difference between their scores for Derivative and Stemitems on the suffix function test) were significantly correlated with standardized measures of generalreading ability only for subjects in seventh grade, and in High School B. The correlation between suffixfunction scores and general reading ability was highest in seventh grade. In High School B, wherescores for several subtests of the Stanford Diarpostic Reading Test were available, the highestcorrelation was with the Word Parts subtest, suggesting that the suffix function score and the word partssubtest both measure some aspect of knowledge of morphology, even though the two tests are quitedifferent in nature.

Stem Discr mination, reflecting the ability to distinguish true roots from pseudoroots was significantlyrelated to standardized ability measures for every group but fourth grade. In seventh grade, there wasan especially strong relationship between Stem Discrimination and a standardized measure of readingcomprehension, r .56, p < .001.

New Word Strategy Questionnaire

Correlational analyses were done to look for relationships between students' answers to the secondquestion (how often the student reported using word parts to figure out the meaning of a new word)and other variables explored in this study. No significant correlations were found between reported useof word parts and any of the experimental or standardized ability measures.

This result, though disappointing, is consistent with Nagy and Scott's (1990) finding that self-reportedreliance on morphology was not significantly related to actual use of morphology in experiments] tasksfor seventh- or tenth-grade students (although they did find a significant correlation in the case ofcollege undergraduates).

1 6


The lack of any significant relationship between reported reliance on morphology and other variablesmay simply reflect the limited validity of this kind of self-report data. On the other hand, this result mayreflect a real lack of relationship, rather than simply a weakness in measurement. Question 2 on thequestionnaire measures a very general level of awareness about word parts, one that is likely to beattained by most students before fourth grade. A student encountering a word such as windowless forthe first time does not necessarily have to reflect on the morphological structure of the word, or on theeffect of the suffix on its part of speech. A student's awareness of the use of word parts may thereforebe determined to a large extent, not by that student's proficiency in using word parts, but by the numberof unfamiliar words the student encounters, and the difficulty the student experiences in using wordparts.

The lack of correlation between reported use of word parts and other measures of morphologicalknowledge and reailing ability suggests that general exhortations to use word parts are unlikely to havemuch impact.

General Discussion

Knowledge of Syntactic Function of Suffixes

One purpose of our study was to get a more accurate picture of students' developing knowledge of thesyntactic function of suffixes. From the results of Tyler and Nagy (1989), it appeared fourth graderswere barely able, if able at all, to demonstrate knowledge of the syntactic function of suffixes in the caseof novel suffixed words. In the present study, using an improved measure, it is clear that fourth gradersare able, to some extent, to determine what a suffix contributes to a novel suffixed word, although thereis also measurable improvement in performance after fourth grade.

On the other hand, although there is some knowledge of the syntactic function of suffixes at all gradelevels tested, this knowledge was not complete; subjects' scores for the derivative versions of items areon the average from 10% to 20% lower than their scores on the stem versions of items. Comparisonof the scores for Derivative and Stem items also allows the identification of subjects who have particulartrouble with suffixes. Hence, this kind of test is potentially a useful diagnostic tool.

A follow-up analysis was performed to explore the possibility that other factors, not directly controlledin the main study, might have contributed to differences in knowledge of the syntactic function of suffix.From the data for High School B, 32 students with the lowest Suffix Function scores were identified,and paired with 32 other students who had high Suffix Function scores. Each pair was matched forperformance on the Stem version, and as far as possible on the Reading Comprehension subscore ofthe Stanford Diagnostic Reading Test. This procedure results in what could be called a High SuffixKnowledge group and a Low Suffix Knowledge group, the two groups being otherwise equivalent. Ateacher familiar with these students was then asked to categorize a randomized list of these 64 studentsby gender, race, and status as a native vs. nonnative speaker of English. None of these factors wasfound to relate significantly to suffix knowledge scores.

Stem Discrimination

Only in fourth grade was there much variation in students' ability to identify the roots of suffixed words.There was more variation in students' tendency to identify pseudoroots as root words.

Performance on the stem discrimination task seems to be most clearly related to the extent to whichthe written instructions were carefully explained to the class. In post hoc exploration of the data, thefourth-grade classes were divided into three groups: One group consisted of the first two classes inwhich the experiment was administered, in which students were individually handed the experimentaltask booklets, briefly informed about the nature of the task, and asked to read the instructions on their

1 7


own. After administering the test to these two classes, it was realized that this level of instruction wouldnot be adequate.

The second group consisted of the bulk of the fourth-grade classrooms, for whom the instructions wereread aloud, with some emphasis and repetition of the most important points. The third group consistedof the two last fourth-grade classes in which the study was conducted, in which the instructions wereboth read aloud and explained, and students were questioned to insure a high level of comprehension.The means for Proportion of Pseudoroots Circled for these three groups were .25, .14, and .10,respectively.

Differences in performance among the high school students tested also appear to indicate thatperformance on the Stem Discrimination task reflected the manner in which the instructions were given.Students in High School A were tested in a single large group in the school cafeteria; they were askedto read the test instructions silently. In High School B, students were tested in classrooms, and theinstructions for the Stem Discrimination task were read out loud and explained to them. Although thetest instructions for the Stem Discrimination task explicitly discussed the difference between roots andpseudoroots and gave examples, it is likely that a substantial proportion of the students in High SchoolA did not pay careful attention to these instructions. When means were computed separately for thetwo high schools, there were significant differences found for both Proportion of Pseudoroots Circledand Stem Discrimination. In fact, the performance of students in High School A was s:milar to that ofthose fourth-grade students who were asked to read over the instructions on their own. It is clear, then,that performance on the Stem Discrimination task is highly sensitive to the manner in which instructionsare given.

We believe, however, that variation in performance on this task may reflect not just failure to followwritten instructions but also some level of misconception about the use of morphology in reading. Thatis, the fact that some students are willing to circle the car in carbon or the pump in pumpkin may be aconsequence of their familiarity with activities in which the object is simply to find words within longerwords or within arbitrary strings of letters without regard to meaning. Such activities may have somevalue, but they certainly run the risk of fostering or confirming misconceptions about morphologicalstructure. Many fourth graders can distinguish between true roots and pseudoroots in iamiliar wordswhen given thorough instructions; but this does not mean that they will keep such a distinction in mindwhen they encounter new words while reading.

Conclusion

Our findings suggest that there is a specific component of morphological knowledge related to, butdistinct from, more general measures of reading ability. In the data for High School B, the twoexperimental measures, Suffix Function Score and Stem Discrimination, correlate more highly with theWord Part subtest of the Stanford Diagnostic Reading Test than witl any other the other subtests, eventhough the Word Part subtest and our two experimental tasks had extremely little in common, in termsof the nature of the test items used, beyond the fact that they all involved morphology in some way.

In general, students were able to perform at high levels on our experimental tasks, demonstrating thatthey possessed, and were able to apply, several types of knowledge about familiar English suffixes. Thus,the inability of students to accurately represent the distinctive contribution of suffixes in their definitionsof derived words observed by Freyd and Baron (1982) and Wysocki and Jenkins (1987) can be attributedlargely to the difficulty of conveying this type of information through verbal defmitions. This result mayalso, at least indirectly, call into question the common practice of teaching the content of derivationalsuffixes via definitions.

On the other hand, there were substantial individual differences in students' knowledge of morphology,which were significantly correlated with standardized measures of reading ability. Our measures allowed

1 8


us to identify students who had a particular deficiency in their knowledge of suffixes. Even in HighSchool B, which showed the best overall performance on the Syntactic Function of Suffixes test, a sizableproportion of students (18%) got at least three fewer of the derivative items correct than of the stemitems, indicating that these students had a substantial gap in their knowledge of suffuces.

Our results also indicated that the ability to distinguish true roots from pseudoroots, although clearlypresent in the fourth-grade students, continued to be significantly correlated with reading ability throughhigh school. Even some high school students showed a tendency to find "little words in big words"despite the lack of semantic relationship between the pseudoroot and the word in which it wasembedded. This indicates a fundamental misconception, either about the nature of word structure ingeneral, or about the nature of the task students thought they were being asked to perform. Suchresults suggest that instruction in morphology must address specific misconceptions or gaps in knowledgeof students at various age and ability levels. The present study constitutes an attempt to develop a testfor the identification of specific difficulties that students might have in dealing with morphologically

complex words.

The study focused only on one type of word structuresuffixationand in particular, only the mostfrequent neutral suffixes. Thus, although the fmdings give us some insight into students' acquisition ofknowledge about English morphology, they cannot be generalized automatically to other kinds ofmorphological structure. Further research is needed to explore students' knowledge of non-neutralsuffixes, as well as prefucation and compounding, and how that knowledge is applied in word learningtasks.

Nagy, Diakidoy, & Anderson Derivational Suffbres - 17

References

Anderson, R. C., & Davison, A. (1988). Conceptual atd empirical bases of readability formulas. In A.Davison & G. Green (Eds.), Linguistic cor.ipleety and tea comprehension: Readability issuesreconsidered (pp. 23-53). Hillsdale, NJ: Erlbaum.

Baker, L. (1989). Developmental change in readers' responses to unknown words. Journal of ReadingBehaviour, 21, 241-260.

Berko, J. (1958). The child's learning of English morphology. Wore4 14, 150-177.

Carroll, J. B. (1940). Knowledge of English roots and au= as related to vocabulary and Latin Study.Journal of Educational Research, 34, 102-111.

Chafe, W., & Danielewicz, J. (1987). Properties of spoken and written language. In R. Horowitz & S.J. Samuels (Eds.), Comprehending oral and written language (pp. 83-113). New York: AcademicPress.

Clark, E. V., Gelman, S., & Lane, N. M. (1985). Compound nouns and category structure in youngchildren. Child Development, 56, 84-94.

Condry, S. (1979). A developmental study of processes of word derivation in elementary school.Unpublished doctoral dissertation, Cornell University, Ithaca, NY.

Davison, A., & Green, G. (1988). Linguistic completely and tea comprehension: Readability issuesreconsidered. Hillsdale, NJ: Erlbaum.

Derwing, B. L., & Baker, W. J. (1979). Recent research on the acquishion of English morphology. InP. Fletcher & M. Garman (Eds.), Language acquisition (pp. 209-223). Cambridge, England:Cambridge University Press.

Freyd, P., & Baron, J. (1982). Individual differences in acquisition of derivational morphology. Journalof Verbal Learning and Verbal Behavior, 21, 282-295.

Gleitman, L R., & Gleitman, H. (1970). Phrase and paraphrase: Some innovative uses of language. NewYork: W. W. Norton & Co.

Graves, M. F. (1986). Vocabulary learning and instruction. Review of Research in Educatiai, 13, 49-90.

Harwood, F. W., & Wright, A. M. (1956). Statistical study of English word formation. Language, 32260-273.

Just, M. A., & Carpenter, P. A. (1987). The psychology of reading and language comprehension. Boston:Allyn & Bacon.

Klare, G. R. (1984). Readability. In P. D. Pearson (Ed.), Handbook of reading research (pp. 681-744).New York: Longman.

Kuhara-Kojma, K., & Hatano, G. (1991, April). Use of a compounding rule in infeffing the meaning ofunfamiliar Kanji compound words. Paper presented at the annual meeting of the AmericanEducational Research Association, Chicago.

20


Mason, J. M., Herman, P. A., & Au, K. H. (1990). Children's developing knowledge of words (Tech. Rep.No. 513). Urbana-Champaign: University of Illinois, Center for the Study of Reading.

Nagy, W., & Anderson, R. C. (1984). How many words are there in printed school English? ReadingResearch Quarterly, 19, 304-330.

Nagy, W., Anderson, R. C., & Herman, P. A. (1987). Learning word meanings from context duringnormal reading. American Educational Research Journal, 24, 237-270.

Nagy, W., & Scott, J. (1990). Word scbemas: Expectations about the form and r leaning of new words.Cognition and Instruction, 7, 105-127.

Selby, S. (1972). The development qf morphological rules in children. British Journal of EducationalPsychology, 42, 293-2'9.

Shepherd, J. S. (1973). The relations between knowledge of word parts and knowledge of derivatives amongcollege freshmen. Unpublished doctoral dissertatior, NeW l'Irk University.

Sternberg, R., & Powell, J. Comprehending verbal comprehension. American Psychologist, 38,878-893.

Tyler, A., & Nagy, W. (1989). The acquisition of English dorivational morphology. Journal of Men ioryand Language, 28, 649-667.

Tyler, A., & Nagy, W. (1990). Use of derivational morphology during reading. Cognition, 36, 17-34.

White, T., Power, M., & White, S. (1989). Morphologi;a1 analysis: Implications for teaching andunderstanding vocabulary growth. Reading Research Quanerly, 24, 283-304.

Winsor, P., Nagy, W., Osborn, J., & O'Flahavan, J. (in preparation). Structural analysis: Toward anevaluation of instructionel practice.

Wysocki, K., & Jenkins, J. (1987). Deriving word meanings through morphotogical generalization.Reading Research Quarterly, 22, 66-81.

Nagy, Diakidoy, & Anderson Derivational Suffices - 19

Table 1

Items Measuring Syntactic Knowledge of Suffixes Used by Tyler & Nagy (1989)

a) Real word Item

You can the effect by turning off the lights.

a) inteusifyb) intensificationc) intensityd) intensive

b) Nonsense stein Item

I wish Dr. Who would just and get it over with.

a) transumpationb) transumpativec) transum pated) transumpatic

.


Table 2

Sample Syntactic Function of Suffix Item

a) Derivative version of the item

Which sentence uses the word powderize most correctly?

a) First they had to fmd a powderize rockb) First they had to powderize find the rockc) First they had to fmd a powderize for the rockd) Ftrst they had to find a way to powderize the rock

b) Stem version of the item

Which sentence uses the word smash most correctly?

a) First they had to fmd a smash rockb) First they had to smash fmd the rockc) First they had to find a smash for the rockd) First they had to find a way to smash the rock


Table 3

Questions from New Word Strateg Questionnaire

1. When you arc reading, how many of the words don't you know?

2. When you come across a word You don't know, how often do you use word parts to figure outwhat the word means?

3. When you come across a word you don't know, how often do you use a dictionary or glossary tofind out what the word means?

4. When you come across a word you don't know, how often do you just slap the word?

5. When you come across a word you don't know, how ofteli do you use the context to figure outwhat the word means?

6. When you come across a word you don't know, how often do you ask someone what the wordmeans?

7. When you come across a word you don't know, how often do you sound out the word to see if youknow what it means?

8. When you come across a word you don't know, how often do you use a dictionary or glossary tofind out how to pronounce the word?

9. When you come across a word you don't know, how often do you try to break the word intosyllables or parts to find out how to pronounce the word?

Nagy, Diakidoy, & Anderson

Table 4

Derivational Suffixes - n

Performance on the Suffix Function Test: Means for Stem and Derivative ItemVersions

StemVersion

DerivativeVersion

SuffixFunction

SCOre

Fourth Grade(n = 210)

.67 (.29) .40 (.34) -.048 (.21)

Seventh Grade(n = 118)

.83 (.25) .62 (.33) -.003 (.20)

High School(n = 311)

.82 (.25) .67 (.33) -.002 (.17)

All Subjectsn .. 639)

.77 (.27) .57 (.35) .000 (.19)

Note: Values for Stem Version and Derivative Version are mean proportion correct, corrected forguessing; standard deviations are in parentheses.

Nagy, Diakidoy, & Andersor

Table SMeans for the Stem Discrimination Test

Derivational Suffbces - 23

Proportion ofTrue Roots

Circled

Proportion ofPseudoroots

Circled

StemDiscrimination

(Relative Proportionof True Roots

Circled)

Fourth Grade(n .. 221)

.83 (.17) .13 (.20) .86 (.15)


.91 (.10) .08 (.14) .93 (.10)

High School(n a 224)

.92 (.10) .15 (.23) .89 (.15)

All Subjects(n is 639)

.88 (.13) .13 (.20) .88 (.14)

Note: Standard deviations arc in parentheses.

Nany, Diakidoy, & Anderson

Table 6

Derivational Suffixes - 24

Correlations Between Standardized Ability Measures and Subjects' Performance onthe Suffix Function Test

StemScore

DerivativeScore

Suffix FunctionScore

,

Fourth Grade(n = 83)

.56s. .40. .16


34*. kw. .34"

High School A(n = 76)

.2b' .35* .17

High School B(n = 140)

Reading Comprehension .49" .53" .28.*

Vocabulary .38" 35" .14

Word Parts .304,* .45" .33"

Structural Analysis .20* 31" .25*

Note: 'Pp < .01 "p < .001

27


Table 7

Derivational Suffutes - 25

Correlations Between Standardized Ability Measures and Subjects' Performance onthe Stem Discrimination Test

Proportion ofTrue Roots

Circled

Proportion ofPseudoroots

CircledStem

Discrimination

Fourth Grade(n a 83)

.32* -.18 .25

Seventh Grade(n a 93)

.17 -.48 .56.

High School A(n a 76)

.13 -33* .40

High School B(n a 140)

Reading Comprehension .05 -.41" 43"

Vocabulary .04 -.38 43**

Word Parts .08 -.45** .48**

Structural Analysis -.04 -.22* 25*

Note: *p < .01 **p < .001

8


Table 8

Derivational Suffixes - 26

Summary of Correlations Among Experimental and Standardized Measures

a) Fourth Grade

SuffixFunction

SrnreStem

Discrimination Ability

Suffix 1.00 .22 .16FunctionScore

Stem .22 1.00 .25Discrimination

Ability .16 .25 1.00

AMMINEL

b) Seventh Grade

SuffixFunaion

ScoreStem


Suffix

OMNI..

1.00 .18 .34FunctionScore

Stem .18 1.00 36.Discrimination

Ability .30* 36 1.00

,

Note: 12 < .01 p < .Gd1

Nag, Diakidoy, & Anderson

Table 8 (Continued)

c) High School A

Derivational Suffices - 27

SunkFunction

ScoreStem


Suffix 1.00 .18 .17

FunctionScore

Stem .18 1.00 .40Discrimination

Ability .17 .40.4, 1.00

d) High School B

SuffixFunction

Score

-------------

SternDiscrimination

..v.,

ReadingComprehension Vocabulazy

WordParts

-

StructuralAnalysis

Suffix 1.00 .30 .21* .14 .33" .25*FunctionScore

StemDiscrimination .30 1.00 .43" .43" .48*. .25*

ReadingComprehension .21* .43" 1.00 .59" .62'''' .19*

Vocabulary.14 .43" .59" 1.00 .62'''' 314-0

Word Parts.33" .48 .62'' .62'' 1.00 .29*.

StructuralAnalysis .25* .25* .19* .31" .29" 1.00

Note: sp < .01 "p < .001

3

Nagy, Diakidoy, /1 Anderson Derivational Suffuces - 28

Figure Caption

Figure 1. Proportion correct on Derivative items as a function of proportion correct on Stem items.

31

Derivative Proportion Correct0.2 0.4 0.6 0.8 1.0

o

el

Di

e*:

oe:

so

I%1. ., 5, 111

e %

. s. 561 if

OS 1 b

% e se _.. Sea se ii. Se

4,-.. se Soo is- P %it lis

sefore Son erores t

e40 so,A

, 4 WS. 110Fireil* e 21, ....ee

S. ID.

If e V es .sPio OP

32

DOCUMENT RESUME ED 334 560 AUTHOR Nagy, William · PDF fileDOCUMENT RESUME ED 334 560 CS 010...

Documents

Transcript of DOCUMENT RESUME ED 334 560 AUTHOR Nagy, William · PDF fileDOCUMENT RESUME ED 334 560 CS 010...